3 * Bacula Director -- backup.c -- responsible for doing backup jobs
5 * Kern Sibbald, March MM
7 * This routine is called as a thread. It may not yet be totally
10 * Basic tasks done here:
11 * Open DB and create records for this job.
12 * Open Message Channel with Storage daemon to tell him a job will be starting.
13 * Open connection with File daemon and pass him commands
15 * When the File daemon finishes the job, update the DB.
21 Copyright (C) 2000, 2001, 2002 Kern Sibbald and John Walker
23 This program is free software; you can redistribute it and/or
24 modify it under the terms of the GNU General Public License as
25 published by the Free Software Foundation; either version 2 of
26 the License, or (at your option) any later version.
28 This program is distributed in the hope that it will be useful,
29 but WITHOUT ANY WARRANTY; without even the implied warranty of
30 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
31 General Public License for more details.
33 You should have received a copy of the GNU General Public
34 License along with this program; if not, write to the Free
35 Software Foundation, Inc., 59 Temple Place - Suite 330, Boston,
44 /* Commands sent to File daemon */
45 static char backupcmd[] = "backup\n";
46 static char storaddr[] = "storage address=%s port=%d\n";
47 static char levelcmd[] = "level = %s%s\n";
49 /* Responses received from File daemon */
50 static char OKbackup[] = "2000 OK backup\n";
51 static char OKstore[] = "2000 OK storage\n";
52 static char OKlevel[] = "2000 OK level\n";
54 /* Forward referenced functions */
55 static void backup_cleanup(JCR *jcr, int TermCode, char *since);
56 static int wait_for_job_termination(JCR *jcr);
58 /* External functions */
61 * Do a backup of the specified FileSet
63 * Returns: 0 on failure
66 int do_backup(JCR *jcr)
68 char since[MAXSTRING];
76 if (!get_or_create_client_record(jcr)) {
77 backup_cleanup(jcr, JS_ErrorTerminated, since);
82 * Get or Create FileSet record
84 memset(&fsr, 0, sizeof(fsr));
85 strcpy(fsr.FileSet, jcr->fileset->hdr.name);
86 if (jcr->fileset->have_MD5) {
87 struct MD5Context md5c;
88 unsigned char signature[16];
89 memcpy(&md5c, &jcr->fileset->md5c, sizeof(md5c));
90 MD5Final(signature, &md5c);
91 bin_to_base64(fsr.MD5, (char *)signature, 16); /* encode 16 bytes */
93 Jmsg(jcr, M_WARNING, 0, _("FileSet MD5 signature not found.\n"));
95 if (!db_create_fileset_record(jcr->db, &fsr)) {
96 Jmsg(jcr, M_ERROR, 0, _("Could not create FileSet record. %s"),
97 db_strerror(jcr->db));
98 backup_cleanup(jcr, JS_ErrorTerminated, since);
101 jcr->jr.FileSetId = fsr.FileSetId;
102 Dmsg2(9, "Created FileSet %s record %d\n", jcr->fileset->hdr.name,
107 * FULL backup job to get the time/date for a
108 * differential or incremental save.
110 jcr->stime = (char *) get_pool_memory(PM_MESSAGE);
113 switch (jcr->level) {
116 /* Look up start time of last job */
118 if (!db_find_job_start_time(jcr->db, &jcr->jr, jcr->stime)) {
119 Jmsg(jcr, M_INFO, 0, _("Last FULL backup time not found. Doing FULL backup.\n"));
121 jcr->jr.Level = L_FULL;
123 strcpy(since, ", since=");
124 strcat(since, jcr->stime);
126 Dmsg1(15, "Last start time = %s\n", jcr->stime);
130 jcr->jr.JobId = jcr->JobId;
131 jcr->jr.StartTime = jcr->start_time;
132 if (!db_update_job_start_record(jcr->db, &jcr->jr)) {
133 Jmsg(jcr, M_ERROR, 0, "%s", db_strerror(jcr->db));
134 backup_cleanup(jcr, JS_ErrorTerminated, since);
138 jcr->fname = (char *) get_pool_memory(PM_FNAME);
140 /* Print Job Start message */
141 Jmsg(jcr, M_INFO, 0, _("Start Backup JobId %d, Job=%s\n"),
142 jcr->JobId, jcr->Job);
145 * Get the Pool record
147 memset(&pr, 0, sizeof(pr));
148 strcpy(pr.Name, jcr->pool->hdr.name);
149 while (!db_get_pool_record(jcr->db, &pr)) { /* get by Name */
150 /* Try to create the pool */
151 if (create_pool(jcr->db, jcr->pool) < 0) {
152 Jmsg(jcr, M_FATAL, 0, _("Pool %s not in database. %s"), pr.Name,
153 db_strerror(jcr->db));
154 backup_cleanup(jcr, JS_ErrorTerminated, since);
157 Jmsg(jcr, M_INFO, 0, _("Pool %s created in database.\n"), pr.Name);
160 jcr->PoolId = pr.PoolId; /****FIXME**** this can go away */
161 jcr->jr.PoolId = pr.PoolId;
164 /* NOTE, THIS IS NOW DONE BY THE STORAGE DAEMON
166 * Find at least one Volume associated with this Pool
167 * It must be marked Append, and be of the correct Media Type
168 * for the storage type.
170 memset(&mr, 0, sizeof(mr));
171 mr.PoolId = pr.PoolId;
172 strcpy(mr.VolStatus, "Append");
173 strcpy(mr.MediaType, jcr->store->media_type);
174 if (!db_find_next_volume(jcr->db, 1, &mr)) {
175 if (!newVolume(jcr)) {
176 Jmsg(jcr, M_FATAL, 0, _("No writable %s media in Pool %s.\n\
177 Please use the Console program to add available Volumes.\n"), mr.MediaType, pr.Name);
178 backup_cleanup(jcr, JS_ErrorTerminated, since);
185 * Open a message channel connection with the Storage
186 * daemon. This is to let him know that our client
187 * will be contacting him for a backup session.
190 Dmsg0(10, "Open connection with storage daemon\n");
191 jcr->JobStatus = JS_Blocked;
193 * Start conversation with Storage daemon
195 if (!connect_to_storage_daemon(jcr, 10, SDConnectTimeout, 1)) {
196 backup_cleanup(jcr, JS_ErrorTerminated, since);
200 * Now start a job with the Storage daemon
202 if (!start_storage_daemon_job(jcr)) {
203 backup_cleanup(jcr, JS_ErrorTerminated, since);
207 * Now start a Storage daemon message thread
209 if (!start_storage_daemon_message_thread(jcr)) {
210 backup_cleanup(jcr, JS_ErrorTerminated, since);
214 Dmsg0(50, "Storage daemon connection OK\n");
216 if (!connect_to_file_daemon(jcr, 10, FDConnectTimeout, 1)) {
217 backup_cleanup(jcr, JS_ErrorTerminated, since);
221 jcr->JobStatus = JS_Running;
222 fd = jcr->file_bsock;
224 if (!send_include_list(jcr)) {
225 backup_cleanup(jcr, JS_ErrorTerminated, since);
229 if (!send_exclude_list(jcr)) {
230 backup_cleanup(jcr, JS_ErrorTerminated, since);
235 * send Storage daemon address to the File daemon
237 if (jcr->store->SDDport == 0) {
238 jcr->store->SDDport = jcr->store->SDport;
240 bnet_fsend(fd, storaddr, jcr->store->address, jcr->store->SDDport);
241 if (!response(fd, OKstore, "Storage")) {
242 backup_cleanup(jcr, JS_ErrorTerminated, since);
247 * Send Level command to File daemon
249 switch (jcr->level) {
251 bnet_fsend(fd, levelcmd, "full", " ");
255 bnet_fsend(fd, levelcmd, "since ", jcr->stime);
256 free_pool_memory(jcr->stime);
261 Emsg1(M_FATAL, 0, _("Unimplemented backup level %d\n"), jcr->level);
262 backup_cleanup(jcr, JS_ErrorTerminated, since);
265 Dmsg1(20, ">filed: %s", fd->msg);
266 if (!response(fd, OKlevel, "Level")) {
267 backup_cleanup(jcr, JS_ErrorTerminated, since);
271 /* Send backup command */
272 bnet_fsend(fd, backupcmd);
273 if (!response(fd, OKbackup, "backup")) {
274 backup_cleanup(jcr, JS_ErrorTerminated, since);
278 /* Pickup Job termination data */
279 stat = wait_for_job_termination(jcr);
280 backup_cleanup(jcr, stat, since);
285 * NOTE! This is no longer really needed as the Storage
286 * daemon now passes this information directly
289 static int wait_for_job_termination(JCR *jcr)
292 BSOCK *fd = jcr->file_bsock;
294 jcr->JobStatus = JS_WaitFD;
295 /* Wait for Client to terminate */
296 while ((n = bget_msg(fd, 0)) > 0 && !job_cancelled(jcr)) {
297 /* get and discard Client output */
299 bnet_sig(fd, BNET_TERMINATE); /* tell Client we are terminating */
301 Jmsg(jcr, M_FATAL, 0, _("<filed: network error during BACKUP command. ERR=%s\n"),
305 /* Now wait for Storage daemon to terminate our message thread */
307 jcr->JobStatus = JS_WaitSD;
308 while (!jcr->msg_thread_done && !job_cancelled(jcr)) {
311 struct timespec timeout;
313 gettimeofday(&tv, &tz);
315 timeout.tv_sec = tv.tv_sec + 10; /* wait 10 seconds */
316 Dmsg0(300, "I'm waiting for message thread termination.\n");
317 pthread_cond_timedwait(&jcr->term_wait, &jcr->mutex, &timeout);
321 return JS_ErrorTerminated;
323 return jcr->SDJobStatus;
327 * Release resources allocated during backup.
329 static void backup_cleanup(JCR *jcr, int TermCode, char *since)
331 char sdt[50], edt[50];
332 char ec1[30], ec2[30], ec3[30];
338 Dmsg0(100, "Enter backup_cleanup()\n");
339 memset(&mr, 0, sizeof(mr));
340 jcr->JobStatus = TermCode;
342 update_job_end_record(jcr); /* update database */
344 if (!db_get_job_record(jcr->db, &jcr->jr)) {
345 Jmsg(jcr, M_WARNING, 0, _("Error getting job record for stats: %s"),
346 db_strerror(jcr->db));
349 strcpy(mr.VolumeName, jcr->VolumeName);
350 if (!db_get_media_record(jcr->db, &mr)) {
351 Jmsg(jcr, M_WARNING, 0, _("Error getting Media record for stats: %s"),
352 db_strerror(jcr->db));
356 msg_type = M_INFO; /* by default INFO message */
359 term_msg = _("Backup OK");
362 case JS_ErrorTerminated:
363 term_msg = _("*** Backup Error ***");
364 msg_type = M_ERROR; /* Generate error message */
365 if (jcr->store_bsock) {
366 bnet_sig(jcr->store_bsock, BNET_TERMINATE);
367 pthread_cancel(jcr->SD_msg_chan);
371 term_msg = _("Backup Cancelled");
372 if (jcr->store_bsock) {
373 bnet_sig(jcr->store_bsock, BNET_TERMINATE);
374 pthread_cancel(jcr->SD_msg_chan);
378 term_msg = term_code;
379 sprintf(term_code, _("Inappropriate term code: %c\n"), TermCode);
382 bstrftime(sdt, sizeof(sdt), jcr->jr.StartTime);
383 bstrftime(edt, sizeof(edt), jcr->jr.EndTime);
384 if (!db_get_job_volume_names(jcr->db, jcr->jr.JobId, jcr->VolumeName)) {
385 jcr->VolumeName[0] = 0; /* none */
388 Jmsg(jcr, msg_type, 0, _("%s\n\
392 Backup Level: %s%s\n\
398 Volume names(s): %s\n\
399 Volume Session Id: %d\n\
400 Volume Session Time: %d\n\
402 Termination: %s\n\n"),
406 jcr->fileset->hdr.name,
407 level_to_str(jcr->level), since,
408 jcr->client->hdr.name,
411 edit_uint64_with_commas(jcr->jr.JobBytes, ec1),
412 edit_uint64_with_commas(jcr->jr.JobFiles, ec2),
416 edit_uint64_with_commas(mr.VolBytes, ec3),
419 Dmsg0(100, "Leave backup_cleanup()\n");